The landscape of graph processing systems is fragmented, with various types of systems, including graph databases, graph computation systems, and GNN systems. However, currently, there is no common file format for efficiently storing and exchanging graph data while maintaining its schema and graph semantics.

GraphAr is designed to address this issue by providing a simple, lightweight format for storing and exchanging graph data. GraphAr is a flexible and adaptable format that can be used across a variety of applications by decoupling the storage of metadata, adjacency, and properties. It is also designed to be efficient by leveraging existing formats such as ORC, Parquet, and CSV as much as possible. This makes it easy to integrate with existing data processing systems and provides a level of familiarity for users.

Furthermore, GraphAr preserves the semantics of graphs and can maintain the schema of property graphs. This means that users can maintain the same graph structure across different systems, ensuring that the graph's semantics are not lost.

We have developed both a C++ library for accessing and creating GraphAr files, and a Spark library for transforming data into and from GraphAr files. With these libraries, it's easy to work with GraphAr files in your existing Spark connectors or C++ projects.

As GraphAr matures, we aim to move the project to a vendor-neutral venue, such as Apache, LF AI & Data. This will ensure that the project continues to be developed in an open and transparent manner, with contributions from a broad community of developers.

To get started with GraphAr, check out our GitHub repository, where you'll find detailed documentation and examples of how to use the libraries. We welcome any feedback or contributions to the project.

Let's make graph data archiving and exchange more efficient with GraphAr!